Neighborhood Feature Pooling for Remote Sensing Image Classification

2025-10-28

AI Summary

To address the limited capability of existing methods in modeling texture features for remote sensing image classification, this paper proposes a Neighborhood Feature Pooling (NFP) module. NFP explicitly captures texture structural relationships within pixel neighborhoods by aggregating cross-channel similarity responses over a local receptive field via lightweight convolutional operations—introducing no additional parameters and incurring negligible computational overhead. The module is architecture-agnostic and can be seamlessly integrated into any CNN backbone. Extensive experiments on multiple benchmark remote sensing datasets—including UC-Merced, AID, and NWPU-RESISC45—demonstrate consistent and significant improvements in classification accuracy across diverse architectures (e.g., ResNet and ViT), with average gains of +1.2–2.8%. These results validate NFP’s effectiveness, strong generalizability, and deployment efficiency.

Technology Category

• Computer Vision: Remote Sensing / Geospatial AI
• Natural Language Processing: Language Grounding & Multi-modal NLP
• Machine Learning: Multimodal Learning

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Web Mining and Content Analysis: Normalization, clustering, classification, and summarization of Web text
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs

Abstract

In this work, we propose neighborhood feature pooling (NFP) as a novel texture feature extraction method for remote sensing image classification. The NFP layer captures relationships between neighboring inputs and efficiently aggregates local similarities across feature dimensions. Implemented using convolutional layers, NFP can be seamlessly integrated into any network. Results comparing the baseline models and the NFP method indicate that NFP consistently improves performance across diverse datasets and architectures while maintaining minimal parameter overhead.

Problem

• Proposes neighborhood feature pooling for texture feature extraction
• Captures relationships between neighboring inputs in remote sensing
• Improves classification performance across diverse datasets and architectures

Innovation

• Neighborhood feature pooling extracts texture features
• It aggregates local similarities across feature dimensions
• Implemented via convolutional layers for seamless integration